- From: Clemm, Geoff <gclemm@rational.com>
- Date: Tue, 19 Jun 2001 23:59:53 -0400
- To: ietf-dav-versioning@w3.org
By "typeless", I meant that you do not declare the type of a variable in a Smalltalk routine. You just apply a method to it, and if that method is not supported by that object, then you get a runtime error. So the analogy is that just as it is bad form in a Smalltalk routine to ask an object "are you of type xxx", it is bad form in a HTTP client to ask a resource "are you of resourcetype xxx". But this is just an analogy, and HTTP is not Smalltalk, so even if we agreed that the analogy is valid, it really wouldn't answer the question, so please forgive my playing analogy games in the first place. And if anyone choses to continue this analogy, I promise not to respond (:-). But if someone has something specific and concrete that you could do by inspecting the value of DAV:resourcetype that you could not do by inspecting the values of DAV:supported-method-set and DAV:supported-live-property-set, I would be interested. Note that I did not see enough support (or any support, for that matter :-) in the working group for removing DAV:activity, DAV:version-history, or DAV:baseline, so I am assuming that they stay unless there is a sudden groundswell of support for taking them out. Cheers, Geoff -----Original Message----- From: Jim Amsden [mailto:jamsden@us.ibm.com] Sent: Tuesday, June 19, 2001 5:25 PM To: ietf-dav-versioning@w3.org Subject: RE: Removing the DAV:activity and DAV:version-history and DAV:ba seline resource type values Geoff says: >>But I do care in general that DeltaV (and in general WebDAV) protocols be designed to encourage good client design, and it seems to me that there is a basic lesson from programming language design (i.e. that typeless languages are more suited to environments where clients and servers get changed) that is being ignored in this drive to extend DAV:resourcetype.<< Just to clarify, Smalltalk in not a typeless language, it is strongly typed, but employes dynamic typing. That is, it determines the type of a class at runtime rather than compile time. Such late binding provides extra flexibility in the language at some cost in performance. In any case, such types are named and described with meta-classes in Smalltalk. Imagine using the Smalltalk class libraries without class names!
Received on Tuesday, 19 June 2001 23:54:15 UTC